Mon. Not. R. Astron. Soc. 000, 1–?? (2015) Printed 20 October 2018 (MN LATEX style file v2.2)

CALIFA Spectroscopy of the Interacting Galaxy NGC 5394 (Arp 84): Starbursts, Enhanced [NII]6584 and Signs of Outflows and Shocks

Nathan Roche1 ?, Andrew Humphrey1, Jean Michel Gomes1, Polychronis Papaderos1, Patricio Lagos1, Sebasti´anF. S´anchez2

1 Instituto de Astrof´ısica e Ciˆenciasdo Espa¸co,Universidade do Porto, CAUP, Rua das Estrelas, 4150-762 Porto, Portugal. 2Instituto de Astronom´ıa,Universidad Nacional Auton´omade M´exico, A.P. 70-264, 04510, M´exico, D.F.

23 June 2015

ABSTRACT We investigate the spiral galaxy NGC 5394, which is strongly interacting with the larger spiral NGC 5395 (the pair is Arp 84), using optical integral-field spectroscopy from the CALIFA (Calar Alto Legacy Integral Field Area) survey. Spatially-resolved equivalent widths, emission-line ratios and kinematics reveal many features related to the interaction, which has reshaped the galaxy. Hα maps (with other diagnostic emission lines) show a concentrated central (r < 1 kpc) starburst and three less luminous star-forming regions (one knot far out in the northern arm), and we estimate −1 the dust-corrected total star-formation rate as 3.39 M yr . However, much of the galaxy, especially the outer tidal arms, has a post-starburst spectrum, evidence of a more extensive episode of star-formation a few ×108 yr ago, triggered by the previous perigalacticon. The [NII]6584/Hα ratio is high in the nucleus, reaching 0.63 at the −3 centre, which we interpret as related to high electron density (ne ' 750 cm from 6717 the [SII] 6731 ratio). We find a central region of strong and blueshifted NaI(5890,5896) absorption, indicative of a starburst-driven outflow from the nucleus at an estimated velocity ∼ 223 km s−1. The CALIFA data also show an annular region at radii 2.25–4 kpc from the nucleus, with elevated ratios of [NII], [OI]6300 etc. to the Balmer lines – this is evidence of shock excitation, which might be the result of interaction-triggered gas inflow. Key words: galaxies: individual: NGC 5394, galaxies:interactions, galaxies: evolu- tion, galaxies:starburst arXiv:1507.07150v2 [astro-ph.GA] 29 Aug 2015

1 INTRODUCTION of spirals, which appear to be destined to merge. This well- known system was included in the Arp (1966) catalog of pe- Mergers of two spiral galaxies are one of the most impor- culiar galaxies under the single name of Arp 84, and studied tant processes in galaxy evolution, and typically occur in a further by Arp (1969). On account of its shape it has also sequence of stages over several hundred Myr, with successive been called the Heron Galaxy, with the larger NGC 5395 be- close passages (perigalactica) of the pair, multiple bursts of ing the ‘body and wings’ and the two-armed, northern NGC star-formation, and complex morphological transformations 5394 the ‘neck, head and beak’; this is well-depicted on the preceding the final coalescence into a single galaxy (which Gran Telescopio Canarias gallery image (Fig 1). could become an elliptical or lenticular). Interacting galax- NGC 5394, a barred spiral seen almost face-on and ies at any stage can be well studied by integral field spec- classed as SB(s)b-pec, appears relatively symmetric, but its troscopy (IFS), which may reveal both the current activity structure is unusual and greatly modified by the interaction. and recent history within each resolution element. Within its compact disk it has a small, bright nucleus and 3 NGC 5394 and NGC 5395 are an interacting close pair arms (double on the west) and beyond the disk (radii r > 13 arcsec out to ∼ 40 arcsec) a distinct and much larger system of two long arms extending north and south, the southern ? [email protected] arm reaching NGC 5395.

c 2015 RAS 2 N.D. Roche, A. Humphrey, J.M. Gomes, P. Papaderos, P. Lagos, S.F. S´anchez

The Arp 84 system is described in some detail by Kauf- man et al. (1999), hereafter K99, who observed in Hα with a Fabry-Perot. NGC 5394 and 5395 have a mass and luminos- ity ratio about 1:4, projected separation 28 kpc, and rotate anticlockwise and clockwise respectively, so that their arms are trailing. The interaction (an orbit approximately in the sky plane) is prograde with respect to the rotation of NGC 5394, but retrograde and inclined for NGC 5395. NGC 5394 has strong Hα emission (indicating star-formation) from the nucleus (2.2×10−13 erg cm−2s−1 from Keel et al. 1985) and more from a region in the SW of the disk (the inner western arm) and two bright spots on the inner and outer north- ern arm. K99’s radio observations give the neutral hydrogen 8 (HI) mass in NGC 5394 as 7.3 × 10 M , and the CO data (tracing molecular hydrogen) show a dense concentration 9 (> 10 M ) of H2 in the nucleus. K99 describe NGC 5394 as ‘post-ocular’ in that it resembles the ‘eye-shaped’ inter- acting galaxies IC 2163 and NGC 2535 but is at a slightly later stage where ‘the eye-shaped oval has evolved into inner spiral arms and most of the gas has fallen into the central region, where it fuels a starburst’. K99 reported some evi- dence for a low-velocity outflow from asymmetry of the Hα line (a possible blueshifted component) and small (∼ 10 km s−1) distortions in the Hα velocity map (NE of the nucleus); unconfirmed but will be investigated here. Figure 1. Gran Telescopio Canarias public gallery im- age of Arp 84, NGC 5394 (north) 5395 (south). From Kaufman et al. (2002), hereafter K02, performed further www.gtc.iac.es/multimedia/media/GTC ARP84.jpg CO observations and comparison with several other pass- bands. They estimate NGC 5394 is inclined 15◦ ± 2 to the sky plane, discuss evidence (e.g. 60µm flux) the Hα flux is object’ with a 8µm flux of 2.626 ± 0.015 mJy, ∼ 1% that reduced by as much as an order of magnitude by internal of the whole galaxy (279.8 ± 3.8 mJy), and in the mid-IR dust extinction, and on this basis estimate the total (dust- it is also detected in [NeII], [NeIII], [SIII] and polycyclic −1 corrected) star-formation rate (SFR) as 6±2 M yr (with aromatic hydrocarbons (PAHs). no evidence of an active galactic nuclei - AGN - contribu- Note that the SFRs estimated from observable quanti- tion). The star-formation closely traces the HI and H2 dis- ties will generally depend on the assumed initial mass func- tributions – all three are concentrated in the central kpc, tion (IMF), e.g. the SFR from K02 assumed a Salpeter IMF, but also the disk is ‘lopsided, with more HI, CO, and Hα whereas equivalent SFR estimates based on the Chabrier or emission coming from the western or southwestern side’, and Kroupa IMFs, with fewer low-mass stars, will be 40–45% there was speculation that this resulted from a west-vs-east lower. Lehmer et al. (2010) estimated a (Kroupa IMF) SFR −1 time lag in the formation of disk arms and subsequent in- of 9.9 M yr for the whole Arp 84 system, simply from the 11 −1 fall of gas to the nucleus. They estimate a molecular/atomic infra-red luminosity of 10 M yr , and found its X-ray ratio M(H2)/M(HI) = 2.5–2.7 (even with a lower CO–H2 luminosity from Chandra data consistent with the typical conversion factor applied for starbursts), very high com- Lx − SFR relation of luminous infra-red galaxies (LIRGs), pared to 0.86 for NGC 5395, which contains much more HI meaning it can be accounted for by X-ray binaries without 10 (> 10 M ), and the average ratio of 0.28 for spiral galaxies requiring AGN. Howell et al. (2010) estimated a higher SFR −1 (Casoli et al. 1998). of 20.99 M yr by including the unobscured component Puerari, Valdez-Guti´errez& Hern´andez-L´opez (2005) traced by far-UV luminosity (GALEX) as well as the infra- studied the morphology of both galaxies in the near-infrared red luminosity and modelling with the Salpeter IMF. (JHK), and fit (in the K-band) a bulge profile with reff = Lanz et al. (2013) combined fluxes in many bands from 3.12 arcsec plus a disk of scale length h = 6.35 arcsec for FUV (GALEX) to FIR (Herschel) for a sample of interact- NGC 5394, which is ‘more compact than normal galaxies’. ing galaxies and fitted the SEDs with a program MAGPHYS, From the disturbed, Cartwheel-like structure of NGC 5395 which assumed a Chabrier IMF and gives a SFR averaged they conclude the system is undergoing a strong interaction, over the past 100 Myr. For NGC 5394 and 5395, they esti- −1 a ‘collision rather than a grazing encounter’. Kaneko et al. mate SFRs of 1.54 and 3.69 M yr , and found some differ- (2013) again study the system in CO and depict the contrast ence in their SEDs, with NGC 5394 having relatively more between the compact, nucleated, H2-rich NGC 5394 and the 12-60µm flux and higher dust temperatures (23.2K in the ring-like and HI-rich 5395. cold component) in comparison to 5395 (with 19.7 K) and Smith et al. (2007) observed the system from 3.6 to the non-interacting control sample (about 19K). This is con- 24µm in a Spitzer survey of Arp-catalog galaxies: most emis- sistent with NGC 5394 experiencing a strong (‘stage 4’), sion from NGC 5394 is from the compact nucleus and the and late-stage interaction, with star-formation more cen- Spitzer images also show the prominent bright spot (Hα trally concentrated than in the more massive companion. 10 source) out in the northern arm. Higdon et al. (2014) de- Stellar masses are estimated as 4.79 and 15.14 × 10 M . scribe this hotspot (‘N1’) as an ‘intergalactic star-forming In this paper we investigate the spatially-resolved prop-

c 2015 RAS, MNRAS 000, 1–?? CALIFA Spectroscopy: Interacting Galaxy NGC 5394 3 erties and history of NGC 5394, especially the effects of ing 3650–4840A˚ with a spectral resolution of 2.3A˚ (FWHM). the interaction, using new integral field spectroscopy from Exposure times are 900 sec on each galaxy with V500 and the Calar Alto Legacy Integral Field Area (CALIFA) survey 3 × 600s with V1200, and the respective 3σ limits for de- (S´anchez et al. 2012; Husemann et al. 2013). CALIFA has al- tection of a narrow emission line are 1.0 and 0.6 × 10−17 ready been the basis for a number of papers on global prop- ergs cm−2s−1. Initial processing of these data generated two erties and on in-depth studies of particular galaxies, with (V500 and V1200) data cubes for each galaxy, which are which we can compare – e.g. the famous ‘playing Mice’ inter- supplied on the CALIFA website (califa.caha.es) calibrated acting pair NGC 4676A/B (Wild et al. 2014), and studies of in units of A˚ and 10−16 erg cm−2s−1A˚−1 and rebinned into metallicity gradients in typical spirals and mergers (S´anchez a grid of 1 arcsec spatial pixels. As the spectrograph fibre di- et al. 2014; Barrera-Ballesteros et al. 2015). We will also ameter is 2.7 arcsec, the effective spatial resolution is lower, compare with models of emission-line ratios in star-forming with FWHM from 3 to 4 arcsec. The cubes are 78 × 73 galaxies (e.g. Kewley et al. 2013) and in shocks (Allen et al. (spatial) ×1877 (spectral) with dλ = 2.0A˚ for V500, and 2008); and simulations of galaxy mergers (e.g. Barnes 2004, 78 × 73 × 1701 with dλ = 0.7A˚ for V1200. Struck et al. 2005, Torrey et al. 2012). In this paper we concentrate mostly on the V500 spec- From the CALIFA catalog, the J2000 co-ordinates of tra which contain the lines of interest. We have analyzed the NGC 5394 are R.A. 13:58:33.19 and Dec +37:27:13.11 and datacubes using a locally developed software pipeline called −1 the heliocentric optical recession velocity is 3457 km (z = PORTO3D (Papaderos et al. 2013, Gomes et al. 2015), which 0.01153); the velocity relative to the local standard of rest effectively separates out the stellar and nebular (emission- (Virgo cluster) is given as a slightly greater 3688 km s−1 line) components of each spatial pixel’s spectrum. The stel- −1 −1 (corresponding to z = 0.01230). For H0 = 70 km s Mpc , lar component is fitted with a set of model stellar spectra, ΩM = 0.27, ΩΛ = 0.73 (assumed throughout) the Virgo with a wide range of ages, using the STARLIGHT package velocity corresponds to a distance modulus 33.64 mag (the (Cid Fernandes et al. 2013). The total stellar mass is esti- 10 value listed in the CALIFA catalog) and an angular diameter mated from the PORTO3D fit as 4.6 × 10 M (for Salpeter distance 52 Mpc, so that 1 arcsec subtends 252 parsec. In IMF). For a list of spectral lines, the program generates 2D Section 2 of this paper we describe the data, in Section 3 maps of emission-line flux, equivalent width, velocity offset the distribution of star-formation activity, in Section 4 and (red or blueshift), and the equivalent widths for any stellar 5 diagnostic line ratios and metallicity, in Section 6 the line- absorption lines (for Balmer lines Hβγδ there may be nar- of-sight kinematics and and Section 7 discuss the history of row emission within broader absorption; as was seen for this the interaction (summarizing in Section 8). galaxy by Arp 1969). Further maps show stellar properties estimated from the model fits, such as mean stellar age and metallicity (mass and luminosity weighted). One line map, for [OI]6300, appeared mildly contaminated by the sky line 2 CALIFA OBSERVATIONS [OI]6364 (de-redshifted to 6291A),˚ so to exclude this a new Our data set for NGC 5394 comes from the first CALIFA flux map was computed summing 6297–6307A˚ only. public data release (DR1; Husemann et al. 2013), which consists of optical datacubes (image × wavelength) for the first 100 (of what will be eventually 600+) galaxies in the survey, now updated with Data Release 2 (Garc´ıa-Benito 3 MAPPING THE STAR-FORMATION et al. 2015). CALIFA observations are obtained with the integral-field spectrograph Potsdam MultiAperture Spec- Figure 2 shows the CALIFA image of NGC 5394 in the con- trophotometer (PMAS)/PPAK mounted on the 3.5m tele- tinuum at 6390–6490A,˚ just blueward of Hα, where it ap- scope at the Calar Alto observatory. The PPAK unit con- pears near-symmetrical, but the emission lines will reveal tains 331 densely packed optical fibres arranged in a hexago- a much more complex picture. The Hα emission line is the nal area of 74 × 65 arcsec, to sample an astronomical object best tracer of the current star-formation rate (SFR). CAL- at 2.7 arcsec per fibre, with a filling factor of 65% due to IFA maps of the Hα flux and equivalent width (EW, here in interfibre gaps. the rest-frame) show (Fig 3) the star-formation is concen- The CALIFA galaxy sample is selected from the Sloan trated in four distinct regions, the nucleus (r < 4 arcsec or Digital Sky Survey (SDSS) DR7 catalog, simply on the basis 1 kpc), a region in the SW of the disk (identified by K99 of redshift (0.005 < z < 0.03) and apparent size (angular as the inner of two western spiral arms) and two compact isophotal diameter 45 < D25 < 80 arcsec), so that each ‘hotspots’ in the inner and outer northern tidal arm (see galaxy fitted within, while filling most of, a single field-of- also Fig 4 of K99). Peak Hα equivalent widths are 52.8, 70.3 view. This means the survey covers a very wide range of and 54.1A˚ for the nucleus, SW disk and inner-N hotspot, morphological type, absolute magnitude, stellar mass, star- with the highest of all, 106.7A˚ for the far-N hotspot, but formation rate, colour, etc. (Walcher et al. 2014). NGC 5395 the nucleus produces by far the greatest Hα luminosity. was not targeted in CALIFA as its apparent size is more than Arp (1969) described an ‘unusually sharply defined twice NGC 5394 (radii RK20 = 34 and 71 arcsec; Kaneko et string of condensations coming out of the companion on al. 2013) and exceeds a single field-of-view, so this analysis the inside of the northern arm’. This ‘string’ is visible on is limited to the more compact galaxy. the GTC image, and the inner-N and far-N Hα-emitting Each galaxy was observed with two spectrograph config- hotspots can be identified with the knots at either end. urations, (i) the V500 grating covering the wavelength range Probably the intermediate knots are recent sites of star- 3745–7500A˚ with a spectral resolution of 6.0A˚ (full-width at formation, which now continues only at the extremities. half-maximum - FWHM), and (ii) the V1200 grating cover- There is lower intensity Hα emission distributed over the

c 2015 RAS, MNRAS 000, 1–?? 4 N.D. Roche, A. Humphrey, J.M. Gomes, P. Papaderos, P. Lagos, S.F. S´anchez

NGC 5394 Log-scale continuum 6390-6490 Ang. NGC 5394 Hα Flux

70 70 -14

-1

60 -16 60 -14.5 -1 pixel -1 pixel 50 A 50 -1 -1 -15 -16.5 s s -2 40 40 -2 -15.5 30 -17 30 erg cm λ

pixels (arcsec) pixels (arcsec) -16 20 20 -17.5 -16.5 10 10 Log Flux erg cm Log Flux F 0 -18 0 -17 0 10 20 30 40 50 60 70 0 10 20 30 40 50 60 70 pixels (arcsec) pixels (arcsec)

Figure 2. NGC5394 red-band continuum map from CALIFA data. North is up and East is left, as in all figures. NGC 5394 Hα Emission Equivalent Width 80 70 whole disk (out to r ' 12 arcsec), but almost none from the 70 tidal arms outside of the two hotspots. 60 To illustrate some of the diversity within this galaxy, 60 50 Fig 4 compares small-aperture CALIFA spectra of the nu- 50 cleus, the far-N hotspot (averaged over the central 5 pixels 40 40 of each) and a region in the southern arm (averaged over 30 14 pixels). Wavelengths 5508–5522A,˚ 6222–6236A˚ and 6285– 30 ˚ pixels (arcsec) 6295A (galaxy restframe) are severely affected by strong sky 20 20 OI lines (observed at 5577.3, 6300.3 and 6363.8A),˚ and are masked out in the plots and model fitting. The nucleus has 10 10 Equivalent Width Angstroms the strong emission lines expected for a starburst, although 0 0 [OIII]5007 is relatively weak and the continuum relatively 0 10 20 30 40 50 60 70 red (i.e. redder than flat in Fλ). The two outer regions are pixels (arcsec) bluer and the N hotspot has strong emission lines, while the S arm is not only passive but may be post-starburst on the basis of strong Balmer absorption lines (EW > 5A).˚ The Figure 3. Map of (log scale) Hα emission-line flux arcsec−2 nucleus has notable differences in emission-line ratios com- (above) and equivalent width in A˚ (below). pared to the hotspot, with higher [NII]6584/Hα and lower [SII]6717/[SII]6731. SFRs locally and for the whole galaxy can be estimated in circular annuli, and we also divide the galaxy into 7 re- by summing the Hα fluxes, but there is evidence (K02) gions, shown on Fig 6. Table 1 gives the observed and dust- the emission lines are substantially attenuated by internal corrected Hα fluxes, summed over each of the 7 regions and dust. For star-forming HII regions the intrinsic flux ratio of for the whole galaxy (for the integrated Hα the estimated Hα/Hβ (when corrected for stellar absorption as here) is ex- extinction is 1.98 magnitudes). Fig 7 shows the integrated 4 pected to be 2.86 (from ‘case B’ models with Te ∼ 10 K), CALIFA spectra of each region (note they have different but the observed ratio is increased in proportion to the dust- areas), which will be model-fitted in a later Section. reddening. Using the extinction law of Calzetti et al. (2000), From the dust-corrected fluxes, Hα luminosities in erg supported by the more recent results of Wild et al. (2011), a s−1 are calculated using the distance modulus 33.64 mag. dust-corrected Hα flux is estimated for each individual pixel A star-formation rate in solar masses (M ) per year can 2.615 −42 as F(Hα)corr ' F(Hα)obs[Hα/2.86Hβ] . Here this dust then be estimated as 4.4 × 10 L(Hα) (e.g. Sobral et correction is only applied for pixels where the statistical er- al. 2014); this conversion factor is for the Chabrier IMF. −1 ror σ(Hα/Hβ) < 1.0, which includes all of the disk and the We estimate 3.39 M yr for the whole galaxy. For the star-forming regions in the northern arm (the other pixels Salpeter IMF (with more low mass stars) the SFR would −42 −1 are included in the Hα flux summations but uncorrected). be 7.9 × 10 L(Hα) (Kennicutt 1998) giving 6.09 M yr , −1 Figure 5 shows a map of the Hα dust extinction, which is in agreement with the K02 estimate 6 ± 2M yr . In com- greatest for the nuclear region (2.32 mag). parison, for the Mice galaxies (each larger and about twice To depict the IFS data (without defeating the object as massive as NGC 5394), Wild et al. (2014) estimated SFRs −1 −1 by too much spatial binning) can be challenging and three 6.2 M yr for NGC 4676A and 2.3 M yr for NGC 4676B methods are used here, 2D maps, radial profiles averaged (correcting for dust using UV and mid-infrared fluxes).

c 2015 RAS, MNRAS 000, 1–?? CALIFA Spectroscopy: Interacting Galaxy NGC 5394 5

[NII] [OIII] [SII] [OI] 6717 Region F(Hα)obs F(Hα)corr L(Hα)corr SFRcorr Hα Hβ Hα Hα [SII] 6731 −2 −1 −2 −1 −1 −1 erg cm s erg cm s erg s M yr Nucleus 1.981 × 10−13 1.681 × 10−12 5.749 × 1041 2.53 0.592 0.273 0.288 0.023 1.069 SW disk 3.776 × 10−14 1.547 × 10−13 5.290 × 1040 0.23 0.452 0.262 0.296 0.031 1.375 Outer disk 9.414 × 10−14 3.603 × 10−13 1.232 × 1041 0.54 0.551 0.407 0.343 0.043 1.351 Inner-N hotspot 5.273 × 10−15 1.154 × 10−14 3.946 × 1039 0.017 0.466 0.425 0.289 0.033 1.301 Far-N hotspot 7.257 × 10−15 2.031 × 10−14 6.946 × 1039 0.031 0.408 0.280 0.270 0.019 1.526 N arm 1.161 × 10−14 1.578 × 10−14 5.396 × 1039 0.024 S arm 8.861 × 10−15 1.088 × 10−14 3.721 × 1039 0.016 Whole Galaxy 3.629 × 10−13 2.242 × 10−12 7.710 × 1041 3.39 0.562 0.368 0.312 0.033 1.191

Table 1. Integrated Hα luminosities for each of 7 regions of the galaxy NGC 5394 (Fig 6); fluxes and luminosities corrected for dust using the Balmer decrement; star-formation rates derived from the corrected L(Hα) (Chabrier IMF); and several important line ratios from integrated, uncorrected fluxes (not shown for tidal arms as the line fluxes are so low).

NCG 5394 (spectra of regions) 14 Nucleus Hα Hα dust extinction (from Hα/Hβ ratio) -2 12 2.5 [NII] 70

arcsec 10 -1

A 60 2 -1

s 8 [SII] -2 50 1.5 6

erg cm Hβ

-16 Hδ Hγ 40 4 [OIII]5007 NaI abs. 10

1 ) in magnitudes λ α pixels (arcsec) 30

2 A(H Flux F 0.5 20 0 4000 4500 5000 5500 6000 6500 Wavelength Angstroms 10 0 10 20 30 40 50 60 70 Far-N Hotspot pixels (arcsec) 0.45 S Tidal Arm (scaled x3) Hα

-2 0.4 Figure 5. Hα dust extinction as estimated from the Balmer 0.35 arcsec Hδ Hγ Hβ decrement Hα/Hβ using the Calzetti et al. (2000) law. -1

A 0.3 [NII] -1 s -2 0.25 [SII] 0.2 erg cm -16 0.15

10 [OIII]5007 λ 0.1 Flux F 0.05

0 4000 4500 5000 5500 6000 6500 Wavelength Angstroms Figure 4. CALIFA V500-grating spectra in 3 different regions of NGC 5394, the nucleus, the far-north star-forming hotspot (av- eraged over central 5 pixels), and the (post-starburst) southern tidal arm.

We estimate the far-N hotspot accounts for 0.9% of the galaxy’s SFR, the same as its fraction of the 8µm flux (2.626/279.8 mJy) in Spitzer observations (Higdon et al. Figure 6. Map of regions the galaxy is divided into for the anal- 2014). Such features are not unexpected e.g. the ‘ocular’ IC ysis in Table 1: (1) S Arm (2) N arm (3) Outer disk (4) Inner-N 2163 has at least 9 star-forming clumps spaced around the hotspot (5) Far-N hotspot (6) SW disk (7) Nucleus ‘eye’, with similar Hα and 8µm fluxes (Kaufman et al. 2012). In NGC 5394, star-formation is now concentrated in the nu- cleus, with 75% at r < 1 kpc. Fig 8 shows the observed and

c 2015 RAS, MNRAS 000, 1–?? 6 N.D. Roche, A. Humphrey, J.M. Gomes, P. Papaderos, P. Lagos, S.F. S´anchez

NCG 5394 (integrated spectra of regions)

Nucleus (r<4 arcsec) Outer disk δ -13.5 SW disk N arm NGC 5394 H Absorption Equivalent Width Far-N hotspot S arm 9 Inner-N hotspot 70 -1

A -14 8 -1 60 s

-2 7 -14.5 50 6 erg cm λ 40 5 -15 4 30 Log Flux F pixels(arcsec) 3 -15.5 20 2

10 Equivalant width in Angstroms -16 1 4000 4500 5000 5500 6000 6500 Wavelength Angstroms 0 0 0 10 20 30 40 50 60 70 Figure 7. CALIFA V500-grating spectra integrated over 7 re- pixels (arcsec) gions of NGC 5394 as shown in Fig 6, on a log scale.

Figure 9. Map of Hδ absorption-line equivalent width. NGC5394: Hα Radial Profile

-2 centred on Nucleus, dust-corrected -13 centred on Nucleus, uncorrected Far-N hotspot, dust-corrected diagnostic of a post-starburst region or galaxy (e.g. Goto arcsec

-1 Far-N hotspot, uncorrected s red continuum, per Angstrom 2008) and is caused by an enhancement in the fraction of -2 -14 A-type stars, which on the basis of stellar lifetimes implies they formed in a period of increased star-formation between -15 100 Myr and 1 Gyr ago. Only in the active star-forming (Hα-emitting) regions is EW(Hδ) reduced to a more typical -16 3–4A,˚ presumably because the A-type spectrum is mixed with even younger OB stars. This shows NGC 5394 is very -17 much a post-starburst galaxy with the earlier burst more spatially extended than the current activity.

surface brightness Log10 erg cm -18 Of much importance in identifying AGN and other α H sources of high excitation, and estimating metallicity, are the 0 5 10 15 20 25 30 35 radius (arcsec) line ratios [NII]6584/Hα (the N2 index) and [OIII]5007/Hβ (the O3 index). These and the other line ratios considered Figure 8. Radial profile of Hα emission, centred on the nucleus, here (except Hα/Hβ) are almost insensitive to dust because as observed and as corrected for dust using the Hα/Hβ ratio. Also shown, at r < 10 arcsec, the profile of the hotspot in the the two wavelengths are very close. For the galaxy as a whole northern arm, and for comparison, the shallower profile of the O3 = 0.368 and N2 = 0.562. Fig 10 and 11 show the ratio red continuum (6390–6490A)˚ in flux A˚−1. maps, and Fig 12 the radial profiles (centred on the nucleus, and for comparison at the far-N hotspot). Many pixels, gen- erally in the outer galaxy where emission lines are weak or dust-corrected Hα flux profile, averaged in circular annuli absent, give very noisy or meaningless ratios, and these maps centred on the nucleus (the peak at r = 29 arcsec is the far- and profiles include only pixels where the ratio measurement N hotspot), and also profiles centred on the far-N hotspot is > 3σ. itself out to r = 10 arcsec only. The nucleus and hotspot The O3 ratio is low at 0.25–0.3 within the main star- are very different in luminosity but are concentrated star- forming regions, with no increase at the centre (only 0.254 formation regions and have similarly steep profiles. In the for the central 4 pixels), so gives no evidence for an AGN. In continuum, the profile of the galaxy is shallower, closer to contrast, N2 is strongly peaked at the nucleus, 0.619 for the an exponential, without the steepening at r ≤ 10 arcsec. The central 4 pixels and 0.629 at the highest pixel, but falls to central star-formation density is very high (higher than in 0.52 at r = 6 arcsec, and 0.41 in the far-N hotspot, which is NGC 4676A), with an integrated Hα flux 1.03 × 10−13 (ob- more typical for star-forming regions. However, at the outer served) or 9.39 × 10−13 (corrected) in the central 14 pixels edge of the disk the mean N2 increases again to > 0.6, and in 2 −1 −2 (0.889 kpc ); the latter corresponds to 1.6 M yr kpc . many pixels here it is even higher than at the centre, with ratios 0.65–0.85. At the same radii there is an increase in O3, and to some extent in [SII]/Hα and [OI]/Hα. We find an outer region of enhanced line ratios forming an irregular 4 LINE RATIOS annulus, mostly between r ' 9 and 16 arcsec. The N2 ratios We examine a number of other diagnostic lines and ratios. in both the nucleus and this outer region exceed the log First, the equivalent width of the Hδ stellar absorption line (N2)=-0.25 upper limit of typical star-formation (e.g. the (Fig 9) is high (6–9A)˚ over most of the galaxy area, espe- Kauffmann et al. 2003 divide). cially the tidal arms. An EWabs(Hδ) > 5A˚ is considered CALIFA data for the Mice galaxies (Wild et al. 2014)

c 2015 RAS, MNRAS 000, 1–?? CALIFA Spectroscopy: Interacting Galaxy NGC 5394 7

NGC 5394 [NII]6584/Hα Flux Ratio NGC 5394 [OI]6300/Hα Flux Ratio 70 0.1 70 0.7 60 60 0.65 0.08 0.6 50 50 α 0.55 0.06 α 40 0.5 40 0.45 0.04

30 [OI]6300/H [NII]6585/H pixels (arcsec) 0.4 pixels (arcsec) 30 20 0.35 0.02 20 10 0.3 0.25 10 0 10 20 30 40 50 60 70 10 20 30 40 50 60 70 pixels (arcsec) pixels (arcsec)

Figure 10. Map of [NII]/Hα ratio (N2 index). Figure 13. Map of [OI]6300/Hα ratio.

NGC 5394 [OIII]5007/Hβ Flux Ratio NGC 5394 [SII]6717+6731/Hα Flux Ratio 70 0.6 0.6 70 60 0.5 60 0.5 α β 50 0.4 50 0.4 40 0.3 40

[OIII]5007/H 0.3 pixels (arcsec) 30 0.2 30 pixels (arcsec) [SII]6717+6731/H 20 0.1 20 0.2

10 0 10 10 20 30 40 50 60 70 0.1 10 20 30 40 50 60 70 pixels (arcsec) pixels (arcsec)

Figure 11. Map of O[III]5007/Hβ ratio (O3 index). Figure 14. Map of the S[II]6717+6731/Hα ratio.

NGC5394: Radial Profiles of N2 and O3 line ratios 0.9 showed large biconical regions in both disks with N2 >

0.8 −0.25 dex (0.56) up to N2 ' 1, with spatially matching high values for the ratios [OI]6300/Hα (with 0.1-0.25) and 0.7 [SII]/Hα (with 0.6-1.0), and (in 4676A) O3 (which ranged from 1–4). These high ratios and the bicone morphology 0.6 could be explained by high-velocity (∼ 350 km s−1) shocks

0.5 with ‘ionized precursor’, from a super wind, an outflow

Flux ratio driven by a central starburst. If the nuclear starburst in 0.4 NGC 5394 causes similar shock ionization, this would ap- pear directly in front of the nucleus due to the face-on ori- 0.3 [NII]/Hα, Nucleus [NII]/Hα, Far-N hotspot entation. However, at the nucleus, O3 and the other two key 0.2 [OIII]/Hβ, Nucleus ratios [OI]6300/Hα and [SII](6717+6731)/Hα (Fig 12 to 15) [OIII]/Hβ, Far-N hotspot do not show high values or any peak relative to larger radii, 0 2 4 6 8 10 12 14 16 radius (arcsec) and all three (in contrast to the N2 ratio), remain within the range expected for HII regions, and so give no evidence Figure 12. Radial profiles of the line ratios N2 and O3 (centred of shocks (or AGN) here. on the nucleus), and for comparison the profiles centred on the The flux ratio of the two lines in the [SII] 6717,6731 far-N hotspot. doublet is a useful diagnostic of electron density (ne) in neb-

c 2015 RAS, MNRAS 000, 1–?? 8 N.D. Roche, A. Humphrey, J.M. Gomes, P. Papaderos, P. Lagos, S.F. S´anchez

NGC5394: Radial Profiles of S2 and [OI]6300/Hα line ratios NGC5394: Radial Profiles of [SII]6717/[SII]6731 (electron density diagnostic)

[SII]/Hα, Nucleus 1.7 0.5 [SII]/Hα, Far-N hotspot [OI]6300/Hα,nucleus [OI]6300/Hα, Far-N hotspot 1.6

0.4 1.5 1.4 ne=50 n =100 0.3 1.3 e

1.2 Flux ratio Flux ratio n =300 0.2 e 1.1

1 ne=600 0.1 0.9 [SII]6717/6731, Nucleus ne=1000 [SII]6717/6731, Far-N hotspot 0 0.8 0 2 4 6 8 10 12 14 16 0 2 4 6 8 10 12 14 radius (arcsec) radius (arcsec) Figure 15. Radial profiles of the line ratios OI/Hα and Figure 17. Radial profiles of the line ratios [SII]6717/[SII]6731 in [SII](6717 + 6731)/Hα in the disk, and centred on the far-N the disk, and centred on the far-N hotspot. An ne scale is shown hotspot. for Te = 10000K.

2 pixels in Nucleus SW disk Far-N hotspot NGC 5394 [SII]6717/[SII]6731 Flux Ratio 1.8 inner-N hotspot 70 1.8

1.7 1.6 60 1.6 1.4 50 1.5 ne=50 ne=100 1.4 1.2 40 [SII]6717/[SII]6731 n =300 1.3 e

pixels (arcsec) 30 1.2 1 ne=600

[SII]6717/[SII]6731 central pixel 1.1 ne=1000 20 0.8 1 0.35 0.4 0.45 0.5 0.55 0.6 0.65 0.7 10 0.9 [NII]/Hα 10 20 30 40 50 60 70 pixels (arcsec) Figure 18. Plot of the line ratios N2 vs. [SII]6717/[SII]6731 in the strongly star-forming regions, showing that they are corre- lated (the correlation coefficient is 0.8375) forming a sequence of Figure 16. Map of [SII]6717/[SII]6731 ratio, sensitive to electron increasing ne. An ne scale is shown for Te = 10000K. density ne

region with 1.30 (ne ∼ 100), to the very centre with 1.07 2 4 −3 −3 ulae, most sensitive at ne ∼ 10 –10 cm (with also a small (ne ' 328 cm ). In the starburst-merger Arp 299 (Heck- dependence on temperature, see e.g. Cant´oet al. 1980). In man et al. 1999) ne similarly increases from ∼ 50 to 250–300 NGC 5394, the [SII]6717/6731 ratio (Fig 16 and 17) shows cm−3 at the centre. Le Tiran et al. (2011) found the stacked a strong monotonic gradient at r < 6 arcsec, falling from spectra of z = 1.3–2.6 starburst galaxies to indicate mean ne 1.4 in the outer disk (r ∼ 6–12 arcsec) to 0.97 at the cen- values of 200 cm−3 (range 100–500) for the Hα-brightest re- tral 5 and 0.949 in the central one pixel. For the calibration gions, similar to local starbursts. On this basis, the ne ∼ 750 −3 in iraf.stsdas.nebula.temden, assuming Te = 10000K, this cm in the NGC 5394 nucleus is a fairly extreme case, prob- −3 signifies a rise in electron density from ne ' 27 cm to a ably a consequence of the interaction. central 750 cm−3. The higher [SII]6717/6731 ratio of 1.55- Kewley et al. (2013) predict in stellar photoionization −3 1.6 in the far-N hotspot indicates a lower density, probably models that increasing ne from 10 to 1000 cm could, by in- −3 ne ≤ 20 cm . For the galaxy as a whole the integrated creasing collisional excitation, raise the N2 ratio by as much [SII]6717/6731 ratio is 1.191 (corresponding to 252 cm−3). as 0.2 dex (their Fig 2; note that O3 is also increased by a In NGC 4676, Wild et al. (2014) found intermediate similar amount). This might in itself account for the central −3 [SII]6717/6731 ratios of 1.1–1.2 (ne ' 200–400 cm ) in high peak of N2 in NGC 5394, even without shocks or AGN. the bicones and both disks, without obvious radial gra- Observationally a high ne is associated with a positive N2 dients. The non-interacting M33 spiral, studied with the offset in the BPT diagram for SDSS galaxies (Brinchmann, same PMAS spectrograph (Lopez-Hernandez et al. 2013), Pettini & Charlot 2008) and for individual HII regions in does show a gradient from an outlying HII region with CALIFA (Fig 2 of S´anchez et al. 2015). Plotting for the in- [SII]6717/6731 ' 1.5–1.8 (very low ne), to the nuclear HII dividual pixels in the 4 main star-forming regions (Fig 18),

c 2015 RAS, MNRAS 000, 1–?? CALIFA Spectroscopy: Interacting Galaxy NGC 5394 9

NCG 5394 (spectra of two outer regions) component extends over the whole disk, but is not apparent 30 inward of r ∼ 4 kpc where star-formation dominates the Hα [NII] Hδ Hγ Hα flux with a lower N2. In e.g., the elliptical NGC 4696, which Hβ 25 has shocks from an ongoing merger but very little star for- -1 [OI]6300

A [SII] mation (Farage et al. 2010), the N2 ratio remains uniformly -1

s 20 close to 2. -2 Similar high line ratios have also been observed in some

15 passive, non-interacting galaxies, e.g. a mean N2 ratio 2.2 in ergs cm the early-type CALIFA sample of Papaderos et al. (2013), -16

10 which may be the result of hard continua from post-AGB λ 10 (Asymptotic Giant Branch) stars, planetary nebulae or ‘hot Flux F 5 low mass evolved stars’, but in these cases the lines are ˚ Annular high-excitation region (296 pixels) fainter, with EW ≤ 2A. Fig 19 shows for comparison a S tidal arm region (scaled to 250 pixels) 0 scaled-up spectrum from a region in the southern tidal tail 4000 4500 5000 5500 6000 6500 (also on Fig 4). The spectra are quite similar except the tail Wavelength (Angstroms) rest-frame is a little bluer (less dust?), has stronger Balmer absorption, Figure 19. Summed spectrum of the annular high-excitation re- but lacks the emission lines (there is possibly a weak [NII] gion where the N2 ratio exceeds 0.65, in total 296 arcsec2. Also line but the EW is no more than 0.8A).˚ This is the part of shown for comparison is the (scaled-up) spectrum from a 14 pixel the galaxy most dominated by the intermediate age stellar region of the S tidal tail. population. These stars cannot be making more than a very small contribution to the high excitation region, and we look elsewhere for the power source (see section 6.3). N2 is closely correlated with the [SII]6717/6731 ratio, which supports ne as the primary cause of the central enhance- ment. 5 METALLICITY GRADIENTS AND THE BPT In contrast, the outer, annular region of high N2 and DIAGRAM other ratios is not associated with any increase in ne. To confirm this region is genuine and investigate further we Metallicities of galaxies or sub-regions can be estimated by a sum the spectra of all 296 pixels at 9 < r < 16 arcsec with great variety of methods. One of the simplest requires only N2 > 0.65 to give the spectrum shown in Fig 19. Hα and O3 and N2 ratios to estimate the nebula gas metallicity, [NII]6584 are well-detected, with a ratio that appears close and can be applied even to high redshifts, with dust having to unity on the plot, but in the integrated line fluxes from little effect, but these ratios are subject to other influences PORTO3D Hα is corrected upward for stellar absorption and (especially AGN) and there are uncertainties in calibration. the resulting mean N2 for these pixels (weighted by flux The O3N2 index, given as log10[O3/N2], is -0.387 at the rather than area) is 0.741. In the same way the other mean centre of NGC 5394 and -0.184 for the galaxy as a whole. ratios are found to be [OIII]/Hβ = 0.844, [SII]/Hα = 0.432 In the widely used PP04 (Pettini & Pagel 2004) calibration, and [OI]/Hα = 0.060. The [SII]6717/6731 ratio is 1.37, giv- metallicity expressed as the logarithmic ratio of oxygen to −3 ing a low ne of 20 cm . The Hα emission and Hδ absorption hydrogen, 12 + log(O/H) = 8.73 − 0.32 × O3N2. Marino et EW are both quite moderate at 4.7A˚ and 4.1A˚ so this is nei- al. (2013), comparing line ratios from the CALIFA survey ther a starburst or a predominantly post-starburst region. with calibrations based on electron temperature, derived a The elevated line ratios in this region resemble (but shallower relation, 12 + log(O/H) = 8.533 − 0.214 × O3N2. are a little more moderate than) the shock-dominated outer A different metallicity measure is obtained from the regions of the late merger NGC 3256 (Rich et al. 2011), the stellar component. Stellar and nebular metallicity may dif- extensive eastern shock region of the colliding NGC 7318 fer, as stars preserve a record of the abundances at the time in Stephan’s Quintet (Rodr´ıguez-Baraset al. 2014) and the of their formation, which may be Gyr ago, and gas may bicones in the Mice. All 4 ratios would fit the models of Rich, move in or out. PORTO3D model fitting generated maps of Kewley & Dopita (2011) for composite shock excitation and luminosity-weighted and mass-weighted stellar metallicity HII regions, where the shock component is ∼ 50% (of the (lwZ and mwZ), the former giving more weight to recently Hα flux). Further comparison with the models tabulated by formed stars, in units of Z (total mass fraction of Natomic ≥ 3 Allen et al. (2008) finds a dominant shock component that elements), which are here converted to O/H on the basis that can fit the observed line ratios quite well with shock velocity solar metallicity is Z = 0.015 and 12 + log(O/H) = 8.72 (Al- v ' 225–275 km s−1 and little or no pre-ionization (which lende Prieto, Lambert & Asplund 2001). Fig 20 shows the would give a higher O3); this assumed near solar metallicity, O3N2 nebular metallicity averaged in radial annuli, using magnetic equipartition (B = 3.23µG) and a low electron only pixels with F(Hα) > 10−16 and signal/noise > 3 for −3 density ne = 1 cm . both ratios (so there is a gap between the disk and the N The total Hα flux (uncorrected for dust) from this shock hotspot). Pixels with N2 > 0.7 or O3 > 0.7 are also excluded region is 9.4 × 10−15 erg cm−2s−1 (L = 3.2 × 1039 erg s−1) on the basis that they are likely to be more dominated by only 2.6% that of the whole galaxy, so shocks remain a small shocks than by star-formation, but this has little effect on (few percent) component of the galaxy’s emission-line lumi- the mean O3N2 values (shocks might cause O/H to be un- nosity compared to star-formation. Note that this region derestimated by increasing the O3 ratio but this may largely only appears annular in ratio maps, while all the line fluxes be compensated by the increase in N2). Fig 20 also shows the increase towards the centre, so it is possible this emission stellar lwZ and mwZ in the same annuli (using all pixels).

c 2015 RAS, MNRAS 000, 1–?? 10 N.D. Roche, A. Humphrey, J.M. Gomes, P. Papaderos, P. Lagos, S.F. S´anchez

NGC 5394: Metallicity radial profile -0.1

8.9 -0.2

8.8 -0.3 )

8.7 β -0.4

8.6 -0.5

8.5 -0.6 central pixel

-0.7 pixels in Nucleus

8.4 Log10([OIII]5007/H SW disk Metallicity 12+log(O/H) Far-N hotspot 8.3 O3N2 PP04 calibration -0.8 inner-N hotspot O3N2 Marino calibration 12+log(O/H)=8.6 (Marino) Mass-weighted Stellar -0.9 12+log(O/H)=8.56 (Marino) 8.2 Luminosity-weighted Stellar Kewley et al.(2013) local sf seq. best-fit Kauffmann et al.(2003) demarcation -1 0 5 10 15 20 25 30 -0.5 -0.45 -0.4 -0.35 -0.3 -0.25 -0.2 -0.15 -0.1 -0.05 radius (arcsec) Log10([NII]/Hα) Figure 20. Radial profiles of the nebular (O3N2) metallicities Figure 21. ‘BPT’ diagram plot of the log of O3 vs. N2, with PP04 and Marino et al. (2013) calibrations (with our fit- for the pixels in the regions of strong star-formation, showing ted gradient), and of the luminosity-weighted and mass-weighted the local star-forming sequence from Kewley et al. (2013), the stellar metallicities from the PORTO3D fits. HII/composite demarcation of Kauffmann et al. (2003) and two iso-metallicity lines based on the Marino et al. (2013) calibration.

The PP04 calibration gives a much higher metallicity than Marino et al. (2013), 12 + log(O/H) ' 8.79/8.85 inte- This was attributed to the inflow of lower metallicity gas. grated/central (1.17–1.35Z ), compared to 8.57/8.62 (0.71– Also gradients in the outer galaxy could be ‘stretched’ out 0.79Z ). The latter is much closer to the model-fit stellar by the formation of tidal tails (Rich et al. 2012). These pro- mwZ. From r = 0.5 to r = 11.5 arcsec the Marino et cesses could explain NGC 5394 having a ‘normal’ O/H gra- al. O3N2 metallicity decreases by 0.080 ± 0.006 dex, but dient at r < 10 arcsec which becomes flat only at r > reff . beyond this remains flat (perhaps even increasing slightly The mwZ shows no significant gradient in the disk (from in the outer hotspot). A linear fit at r < 12 arcsec gives r = 0 to r = 8.5 arcsec changing by −0.0018 ± 0.0331), 8.6166−0.0069(±0.0002)r (plotted on Fig 20), and for r < 8 whereas the lwZ falls steeply (−0.1135 ± 0.0151 between arcsec only, 8.6205 − 0.0083(±0.0003)r. These gradients are r = 0.5 and r = 6.5), and remains much lower than the -0.028 and -0.033 dex/kpc, or with the disk exponential mwZ in the outer disk, e.g. by 0.1359 ± 0.0222 at r = 8.5 (a scale h = 6.5 arcsec (Puerari et al 2005) corresponding to 6σ difference). This is surprising in view of the expectation reff = 10.9 arcsec (2.75 kpc), can also be expressed as -0.076 that metallicity would increase with time, which would give and -0.091 dex/reff . lwZ > mwZ. Again a possible explanation is that the inter- S´anchez et al. (2014) found CALIFA disk galaxies typ- mediate age and recent starbursts were fuelled by inflowing ically had a metallicity gradient of -0.1 dex/reff at 0.3 < gas of lower metallicity compared to the underlying older r < 2 reff , with flattening at even larger radii and evidence disk stars, and, as Fig 8 of Barrera-Ballesteros et al. (2015) of a shallower mean gradient (-0.05) for interacting than for suggests, the resulting O/H reduction is most apparent not non-interacting (-0.11) subsamples (these estimates used the at the centre but at r ∼ 5–10 arcsec (0.5–1 reff ). PP04 calibration and would be reduced by 1/3 for that of Returning to the nebular metallicity, Fig 21 shows the Marino et al.). Rich et al. (2012) found similar gradients, BPT (Baldwin, Phillips & Terlevich 1981) diagram, i.e. log which flattened as mergers progressed. NGC 5394 appears O3 vs. log N2, for the principal star-forming regions, to- consistent with their mean for early-stage mergers, but also gether with two isometallicity lines (using the Marino et with the non-interacting disks of the same mass in Ho et al. al. calibration), and the ‘local star-forming sequence’ from (2015). Simulations predict that interactions reduce the cen- Kewley et al. (2013), which is obtained from a photoion- tral O/H and flatten gradients by the infall of low-metallicity ization model (with low electron density) and fits the lo- gas (Rupke, Kewley & Barnes 2010), but there is also a com- cus of typical z < 0.1 star-forming galaxies in the SDSS. peting effect of re-enrichment by star-formation (Torrey et The latter is essentially a sequence of increasing metallic- al. 2012), so the resulting evolution is more complex (in disk ity (right/downwards) and approximately orthogonal to the galaxies with high gas fractions the central O/H may even isometallicity lines [on which ∆(log N2) ' ∆(log O3)]. The be increased). distribution of CALIFA pixels shows the increase in metal- A new study by Barrera-Ballesteros et al. (2015), com- licity from the outer star-forming regions to the central pixel, paring interacting and isolated disk galaxies in CALIFA, but combined with this, a large shift in the orthogonal direc- suggests that the effect of interactions on O/H is more tion, along the isometallicity lines and away from the local complex than a simple progressive flattening. They found sequence, by as much as 0.2 dex in N2. This latter cannot be interactions had little effect on the average central O/H, due to metallicity but is as expected for a steep increase in and might even steepen metallicity gradients at small radii ne towards the centre (Kewley et al. 2013) and we propose (r < 0.5 reff ) where interacting galaxies show an enhance- this is the cause. The entire nuclear region transgresses the ment in SFR. At larger radii (r ∼ 1–2 reff ) the mean metal- Kauffmann et al. (2003) demarcation, which was intended licity is reduced and it must be here the gradient is flattened. to divide pure star-forming galaxies from HII/AGN com-

c 2015 RAS, MNRAS 000, 1–?? CALIFA Spectroscopy: Interacting Galaxy NGC 5394 11

Hα ∆(v) km s-1 [NII]6584 ∆(v) km s-1 Stellar ∆(v) km s-1 (V500 data) Stellar ∆(v) km s-1 (V1200 data) 70 80 70 80 70 80 70 120

60 60 60 60 60 60 60 100 40 40 40 80 50 50 50 50 20 20 20 60 40 40 40 40 0 0 0 40 30 30 30 30 -20 -20 -20 20 pixels (arcsec) 20 20 pixels (arcsec) 20 20 -40 -40 -40 0

10 -60 10 -60 10 -60 10 -20

0 -80 0 -80 0 -80 0 -40 20 30 40 50 60 20 30 40 50 60 20 30 40 50 60 20 30 40 50 60 pixels (arcsec) pixels (arcsec) pixels (arcsec) pixels (arcsec)

Figure 22. Maps of the line-of-sight relative velocity ∆(v), from Figure 24. Map of the stellar line-of-sight relative velocity ∆(v) the red/blue-shifting of the emission lines Hα (left) and [NII]6584 measured by fitting to the spectra (i.e. from absorption lines) on (pixels shown white, which have little emission, do not have valid the V500 and the V1200 data. measurements). Note they are closely similar.

NGC5394: Stellar line-of-sight Velocity Curve for PA=26 deg NGC5394: Emission-line Velocity Curve for PA=26 deg 120 V500 data 80 Nucleus α V1200 data H Nucleus 100 [NII] 60 80

60

40 -1

-1 40

20 (v) km s

∆ 20 (v) km s ∆ 0 0

-20 -20 -40

-40 -30 -20 -10 0 10 20 30 -30 -20 -10 0 10 20 30 radius (arcsec) radius (arcsec) Figure 25. Line-of-sight velocity curve from the stellar spectra, Figure 23. Line-of-sight velocity curve from the Hα wavelength shown for both the V500 and V1200 data averaged in ±15◦ sectors ◦ ◦ averaged in ±15 sectors on the kinematically determined axis, on the position angle φk = 26 . ◦ position angle φk = 26 (East of North), with the NNE direction denoted positive. centred on the nucleus, in 2-arcsec bins of radius and within ◦ ◦ posites, but is too conservative in not fully allowing for the position angles θ = φk ±15 and θ = (φk +π)±15 where φk is the kinematic position angle of the galaxy. The velocity potential [NII] enhancement of high ne and hard radiation fields, which can move HII regions across the line (e.g. Fig curve is evaluated in this way for all PAs φ and φk esti- mated as the angle which maximises the velocity difference 2 of S´anchez et al. 2015). r=22” between the opposite sides, i.e. Σr=0 [vrot(r) − vrot(−r)]; ◦ this gave φk = 26±13 , in agreement with the kinematic (re- ceding side) PA of 25.8◦ measured for this galaxy by Garc´ıa- 6 KINEMATICS Lorenzo et al. (2015). We take φ as 26◦ and Fig 23 shows the Hα v (r) along 6.1 Rotation of the Disk k rot this axis (NNE is positive, SSW is negative). The velocity Fig 22 shows a map of the line-of-sight velocity as measured curve shows the gradient produced by the galaxy’s rotation, from the Hα line, and also for [NII]6584. The galaxy is in- with an amplitude (taken as half the difference between the clined only 15◦ from face-on, according to K02, so that the mean of the two largest velocity measurements on each side) line-of-sight ∆(v) is lower than the true rotation of the disk 35 km s−1. It is relatively symmetric, but with minima on by a factor 3.9. Even so, the velocity gradient from rotation both sides at r ' 10 arcsec followed by steeper gradients, a is clearly visible in the Hα velocity map. We obtain a velocity feature seen in some barred galaxies and which might corre- curve vrot(r) by averaging this map in two opposite sectors spond to the weak distortion K99 found in the Hα velocity

c 2015 RAS, MNRAS 000, 1–?? 12 N.D. Roche, A. Humphrey, J.M. Gomes, P. Papaderos, P. Lagos, S.F. S´anchez map some 8 arcsec NE of the nucleus. The [NII] velocity the NaI line, which is here λ = 5888.45 ± 0.09A,˚ and by −1 curve is very similar and gives an amplitude 43 km s . The comparison with the (STARLIGHT-model) stellar NaI gives disk rotation velocity must be at least ∼ 40 × 3.9 = 156 km the blueshift of the non-stellar component as 4.38 ± 0.26A˚ s−1, which is typical for a spiral galaxy in this stellar mass or 223 ± 13 km s−1. 10 range (∼ 2.4–4.6 × 10 M ), e.g. Cortese et al. (2014). This is indicative of a neutral gas outflow from the Shocks and ionized outflows are associated with mergers nucleus, directed perpendicular to the disk. In Fig 7, the (Rich et al. 2011; Rich, Kewley & Dopita 2014), and might spectrum of the nucleus region, integrated over 50 pix- be revealed by kinematics, e.g. Yuan et al. (2012) attributed els (r < 4 arcsec), shows enhanced NaI absorption with a high-N2 region to shocks/outflows as it was ∼ 100 km EW = 5.52±0.02A˚ at λ = 5890.12±0.02A˚ (almost as strong s−1 blueshifted with respect to the galaxy. The ∆(v) map as for the central 5 pixels only), but elsewhere the NaI line for [NII]6584 might be more sensitive to shocks than Hα. is consistent with the stellar models in EW (only 1–2.5A)˚ However, the two velocity maps are very similar, and neither and λ(' 5892–3A).˚ This confirms the strong blueshifted ab- shows any region of blueshifted emission (ionized outflowing sorption originates in the galaxy centre and is not a spurious gas), either at the nucleus or in the outer (annular) high-N2 feature produced by sky lines or some instrument problem. region. The outflow appears to have a similar, but weaker effect on Fig 24 shows the stellar ∆(v) maps, obtained by fit- the CaII ‘K’ absorption line at 3933A;˚ in the central 5 pixels ting the emission-subtracted spectra with STARLIGHT stel- the observed and stellar model wavelengths are measured as lar models, for both the V500 data and the higher-resolution 3931.48A˚ (EW 3.6A),˚ 3933.04A˚ (EW 2.0A),˚ and the resid- V1200, and Fig 25 the stellar velocity curves measured in ual (excess absorption) is at 3930.30A,˚ which corresponds to ◦ −1 the same way on the PA φk = 26 , centring on the nucleus a very similar outflow blueshift ∆(v) = −209 km s . This in each case. The stellar ∆(v) maps, especially from V500, outflow might also explain the stellar velocity maps appear- show a localized blueshift (∼ 20 km s−1) near r ' 0 and the ing slightly blueshifted at the centre. These ∆(v) could have cause is discussed below. The V500 stellar velocities show a been locally biased (up to ∼ 20 km s−1 blueward) by the weaker velocity gradient, with an amplitude (excluding the effect of the outflow on some absorption lines, especially in centre) only 21 km s−1, but those from the V1200 data have the V500 spectra which contain the NaI doublet. a stronger gradient and a positive offset with maximum am- We calculate (integrating over wavelength) a map of plitude 49 km s−1. Cortese et al. (2014) found that stellar ro- the NaI EW (Fig 27); the regions of excess absorption ap- tation velocities in SAMI (Sydney-Australian-Astronomical- proximately fills the central r < 4 arcsec (r < 1 kpc) Observatory Multi-object Integral-Field Spectrograph sur- and, measuring with IRAF ellipse, has a FWHM 6.4–7.0 vey) disk galaxies averaged 0.14 (±0.11) dex lower than vrot arcsec (much larger than the instrumental FWHM of 3.5 from nebular emission lines, but considering the large un- arcsec) and in the outer isophotes is slightly extended to certainties and low inclination it is not clear if the two differ the SE (at the EW = 3.5A˚ isophote, PA = −36 ± 3 deg significantly in this galaxy. and ellipticity= 0.24 ± 0.02; using IRAF ellipse). Alongside we show a ∆(v) map obtained by Gaussian-fitting the NaI absorption in each pixel (with IRAF fitprof, over 5879 to 6.2 Evidence of an Outflow 5920A,˚ avoiding the weak HeI emission line at 5876A).˚ The We did not see evidence of outflowing ionized gas in the Hα blueshifted and high-EW regions correspond spatially, being and [NII] velocity maps. Also both line profiles appear quite extended by > 1.5 kpc on a PA close to the inner-disk minor symmetric (although asymmetry was reported by K99), at axis, which seems consistent with an outflow perpendicular least at the resolution of the V500 data, and are well-fitted to the disk. Over this central region the total flux absorbed by Gaussians. Any blueshifted line emission must be very in the residual (non-stellar i.e. outflow) line is 1.44 × 10−14 faint compared with the star formation is the disk. How- erg cm−2s−1. ever, the cool gas phase of outflows may instead show itself The EW and ∆(v) of the NaI absorption are compa- in absorption lines, in particular MgII(2796,2803) and NaI rable to the NaI lines in many of the starbursting, LIR = 10−12 (5890,5896). The nuclear spectrum plotted in Fig 4 has a 10 L galaxies in Heckman et al. (2000), which have very strong absorption line at 5890A˚ restframe, which can dust-rich super-winds. Blueshifted NaI absorption is ob- be identified as the NaI (5890,5896) doublet, unresolved by served to a lesser degree in more typical disk galaxies (Chen the V500 grating and slightly blueshifted. While this line et al. 2010); the detection probability increases with central appears in purely stellar spectra (especially from K type SFR, face-on orientation and high dust reddening, and the stars), a stronger and blueshifted NaI absorption, seen in correlation of ∆(v) with disk orientation implies the outflows many LIRGs (e.g. Heckman et al. 2000), is a signature of are perpendicular and entrained within bicones of opening starburst-driven outflows. angles ≤ 60 degrees (Wild et al. 2014 measured this angle To investigate this we sum the stellar model fitted by directly for NGC 4676A as Ωw/4π = 0.37, corresponding to ◦ PORTO3D over the same 5 pixels of the nucleus and compare 51 ). The NaI blueshift/outflow velocity is also correlated −1 −1 this with the observed spectrum (Fig 26). The NaI line ob- with SFR, with ∼ 200 km s for 10 M yr , and Bordoloi served here with λ = 5889.62 ± 0.09A˚ and EW = 7.03 ± 0.06 et al. (2014) found the same correlations for z ∼ 1 galaxy is > 3 times stronger than, and blueshifted with respect to, outflows observed in MgII 2796,2803. Our ∆(v) is slightly the purely stellar model where the blended doublet is fit- above average for the SFR but within the typical range. ted by a Gaussian with λ = 5892.73 ± 0.24A˚ with EW = Heckman et al. (2000) give a formula (eq 5) for estimat- 21 −2 2.26 ± 0.07A.˚ The observed line blueshift is 3.11 ± 0.26A˚ or ing outflow rate as dM/dt ∼ 60(r∗/kpc)(NH/3×10 cm )× −1 −1 −1 158±13 km s . Taking the (observed-model) residual spec- (∆v/200 kms ) × (Ωw /4π)M yr . Putting in starburst 20 trum isolates the non-stellar, interstellar-gas component of radius r∗ = 0.75 kpc, NH = 6.1 × 10 at the centre (from

c 2015 RAS, MNRAS 000, 1–?? CALIFA Spectroscopy: Interacting Galaxy NGC 5394 13 -2 5 6.3 Evidence of Inflow? (a) Nucleus: Observed Spectrum Nucleus: Starlight model fit arcsec 4 The outer high-excitation region (at 9 < r < 16 arcsec) does

-1 Residual (non-stellar), offset +0.7 in flux

A not show the same blueshifted ∆(v) as the outflow from the -1 3 s

-2 nucleus. Although very apparent in line ratio maps it seems 2 to leave no imprint on the velocity maps in either Hα or

erg cm 1 [NII], suggesting little motion perpendicular to the disk of -16 the underlying galaxy. Nor is it associated with a signifi-

10 0 λ

F 4000 4500 5000 5500 6000 6500 cant broadening of the [NII] line (at least at this resolution), Wavelength, Angstroms (rest-frame) which remains at FWHM ' 6A˚ across the galaxy. This may 2.5 2 (c) CaII 3933 (b) HeI 5876 NaI(5890,5896) be similar to the late-stage merger NGC 3256, in which the 2 1.5 outer regions with high, shock-like line ratios are blueshifted 1.5 −1 1 only ‘a few 10 km s ’ (Rich et al. 2011), whereas the central 1 strong NaI absorption (caused by outflow) is blueshifted by 0.5 0.5 ∆(v) = −309 km s−1 (Heckman et al. 2000). 0 0 The non-visibility of the outer high-excitation region in 5750 5800 5850 5900 5950 6000 3900 3950 ∆(v) maps would argue that it is not outflowing gas driven Figure 26. Observed spectrum of the central 5 pixels compared by the nuclear starburst. We suggest that the starburst- with the STARLIGHT stellar model fit, for (a) the whole spectrum driven outflow remains collimated in front of the central and (b) for the region of the NaI(5890,5896) absorption doublet, disk, as on the NaI map (Fig 27), and the outer high ex- together with the observed - model residual, showing that there is substantial blueshifted, non-stellar NaI absorption at the central citation region is powered instead by the kinetic energy of region of NGC 5394. (c) This blueshifted absorption may also be gas spiralling inwards within the disk plane, where it would visible, less strongly, in the CaII (K) line at 3933A.˚ have little ∆(v) in the line-of-sight direction, as a result of tidal forces from the interaction. The gas would lose angular momentum in the shocks and eventually form stars. This mechanism was argued for the outer regions of high excitation/line ratios found in a number of interact- ing or merging LIRGs (Monreal-Ibero et al. 2010, Rich et NaI Absorption EW (Angstroms) NaI Absorption ∆(v) km s-1 al. 2011). One argument was that the occurrence of outer 70 70 high-excitation regions was more correlated with (late) in- 7 150 teraction stages than with high SFR. In the Torrey et al. 60 60 (2012) merger simulations for a similar mass galaxy, the nu- 6 100 clear gas inflow rate is generally ∼ 0.5–1.0 × SFR and 1–5 −1 50 50 M yr . At the velocities of disk rotation ∼ 160–170 km 5 50 s−1 or as estimated for the shocks, ∼ 250 km s−1, and an 40 4 40 −1 0 inflow rate of at a few M yr , the kinetic energy input 3 would be sufficient to power the shock component of line 30 30 -50 40 −1

pixels (arcsec) emission (∼ 10 erg s ). 2 20 20 -100 1 10 10 -150 0 7 HISTORY OF THE INTERACTION 20 30 40 50 20 30 40 50 pixels (arcsec) pixels (arcsec) NGC 5394 is now undergoing a nuclear starburst (generat- ing an outflow), has other localized knots of star-formation, and shows evidence of a more extensive (over most of the 8 9 Figure 27. (a) Map of the equivalent width of the NaI(5890,5896) galaxy) episode of star-formation 10 to 10 years ago. Fig absorption doublet, obtained by integrating over the line (5879– 28 shows the the PORTO3D model-fits mass distributions of 5904A˚), clearly showing the central region of strong absorption young (< 100 Myr) and intermediate age (0.1-1.0 Gyr) stars. ˚ (up to EW = 7A), in excess of that expected for stars alone (about The young population has a total mass 206 million M and 2A);˚ (b) map of the velocity shift ∆(v) of the NaI absorption line is very centrally concentrated with 132 million M in the ˚ relative to the stellar-fit 5892.73A, as obtained by a Gaussian fit central < 4 arcsec and only 1.7 million in the far-N hotspot. to each pixel over 5879–5920A.˚ The dust-corrected Hα SFR in Table 1, especially if fur- ther corrected to the Salpeter IMF of these masses (giving −1 6 M yr ) implies the current SFR is significantly higher −1 8 than the mean (2.06 M yr ) for the past 10 yr. Similarly −1 −1 Fig 5 of K02), ∆v = 223 km s and Ωw/4π ' 0.37 , gives our SFR of 3.39 M is more than twice the 1.54 M yr −1 dM/dt ' 3.77 M yr , similar to our estimate for the SFR. estimate of Lanz et al. (2013) based on SED-fitting with Heckman et al. (2000, and similarly Bordoloi et al. 2014) a model of the same IMF and 100 Myr of constant star- concluded that typically a ‘superwind is expelling matter formation. Hα traces only the very recent SFR (≤ 10 Myr), at a rate comparable to the star formation rate’. The ki- compared to longer timescales for the infra-red emission, so 1 dM 2 40 −1 netic power would be 2 dt v = 5.9 × 10 erg s ; 1/13 the the implication is again the SFR has increased strongly over galaxy’s dust-corrected Hα luminosity (Table 1). the past 108 yr.

c 2015 RAS, MNRAS 000, 1–?? 14 N.D. Roche, A. Humphrey, J.M. Gomes, P. Papaderos, P. Lagos, S.F. S´anchez

two hotspots are not seen in the intermediate-age map, as they are younger features. Mass Density of young, <100 Myr age stars K99/K02 describe NGC 5394 as ‘post-ocular’ and sim- 8 ilar to the interacting galaxy IC 2163 but at a later stage 70 of evolution, with star-formation more concentrated in the 60 7 nucleus. IC 2163 is interacting with the larger spiral NGC 2207 and indeed appears eye-shaped with most of its star- 50 -2 6 formation in numerous hotspots forming two arcs around 40 the rim of the ‘eye’ (Kaufman et al. 2012). In the Struck et arcsec 5 al. (2005) simulation of this pair there was a major starburst 30 Solar at the first close passage, some 280 Myr earlier, and there

pixels (arcsec) 4 is predicted to be a second burst at least 100 Myr in the 20 log M future as the galaxies finally merge. NGC 5394 could be 50– 10 3 100 Myr ahead of IC 2163 and entering this stage, with the SFR increasing again as the galaxy approaches merger. In 0 0 10 20 30 40 50 60 70 this simulation the long tidal arms are drawn out of the ini- pixels (arcsec) tially round galaxy very soon after the first close encounter and starburst, and this might account for the tidal arms in NGC 5394 having the highest content of intermediate-age stars. The Arp 84 system is certainly at a later evolution- Mass Density of 0.1-1.0 Gyr age stars ary/merger stage than the Mice, for which the simulations 8 70 of Wild et al. (2014) put the first perigalacticon only 170 Myr ago, and observationally the oldest star clusters in both 60 7 galaxies date from exactly that time (Chien et al. 2007). The

-2 Mice have less Hδ absorption and much less of an intermedi- 50 6 ate age population than NGC 5394, and in the simulations 40 of Wild et al. (2014) and Barnes (2004) their biggest star- arcsec 5 burst lies 0.4–0.8 Gyr in the future when the galaxies will 30 Solar −1 merge and the SFR peak at 40–50M yr . pixels (arcsec) 4 20 log M The hotspots in the northern arm of NGC 5394 are examples of the ‘hinge clumps’ seen in many interacting 10 3 galaxies (Smith et al. 2014) – bright knots of active star- 0 formation, typically found where tidal arms/tails join to the 40 −1 0 10 20 30 40 50 60 70 disk, with EW(Hα) ∼ 100A˚ and L(Hα) ∼ 10 erg s . An- pixels (arcsec) alytical and numerical models of tidal disturbance (Struck and Smith 2012, SS12) can explain many of the galaxy’s features including the formation of long, curved and paired Figure 28. Distribution of the young (age zero to 100 Myr) and tidal arms (which can extend to ∼ 3 times the radius of the intermediate age stellar population (age 100 Myr to 1 Gyr), as pre-interaction galaxy) and an inner disk with a structure estimated from the STARLIGHT model fits, shown on a log scale described as ocular with internal arms (e.g. Figs 7 and 10 of stellar mass per pixel. of SS12); furthermore the model predicts ‘swallowtail caus- tics’ at the base of the tidal tails or arms where intersect- ing streams of gas would pile up, giving rise to localised The intermediate-age population has a very different star-formation which could be quite sustained and produce distribution, in the outer disk as a broad ring and, especially, ‘hinge clumps’ such as the inner-N hotspot. A refinement of throughout the outer tidal arms. The earlier starburst must the SS12 model, where a prograde encounter is represented have been spread over the whole galaxy, whatever its shape by successive impulses, can produce additional swallowtail at that time. The total mass of intermediate age stars is es- caustics far out from the disk on the spiral arms (Fig 12 of 9 timated as 1.200×10 M with only 170 million M at r < 4 SS12) and in this way account for the formation and posi- arcsec. These patterns recall the merger simulation of Re- tion of the far-North hotspot (and the similar feature in Arp naud, Bournaud & Duc (2015), in which early star-formation 82). Introducing a little asymmetry further improves the re- is extensive but the final burst is ‘almost exclusively central, semblance to real Arp (1966) galaxies. The high-resolution due to gas inflows’. The concentration of star-formation in simulation of Renaud et al. (2015) also predicts the forma- the SW disk (the inner western arm, see K02), reflects the tion of a ‘young massive cluster’ far out on the northern tidal lopsided distribution of H2 which would be a short-term ef- arm. Probably these hotspots will become globular clusters. fect of the interaction, so is seen in the young population In a further attempt to reconstruct the star-formation map but not for the intermediate-age stars, which instead history we run STARLIGHT model fits on the 7 region spectra are concentrated in the other two disk arms (eastern and (Fig 7), with a set of Chabrier-IMF stellar templates repre- outer-western). This agrees with to the ‘west-vs.-east time senting 23 ages from 1 Myr to 13 Gyr and four metallicities lag’ in star- and arm-formation (K99), with only the inner- (Z=0.004, 0.008, 0.02, 0.05). For each fλ(age, Z) template western ‘third arm’ that is delayed i.e. recent. Similarly the spectrum the program fits the mass of stars present now and

c 2015 RAS, MNRAS 000, 1–?? CALIFA Spectroscopy: Interacting Galaxy NGC 5394 15

Starlight model star-formation histories for regions of NGC 5394 rich galaxies (Monreal-Ibero et al. 2010, Rich et al. 2011, 3.5 Nucleus 2014; Rodr´ıguez-Baraset al. 2014). Outer Disk

-1 3 N arm We also find evidence for a nuclear-starburst-driven out- yr 2.5 S arm −1 2 flow of gas with (blueshift) velocity 223 ± 13 km s , and Solar −1 1.5 estimated the rate as ∼ 3.77 M yr , which from Heckman 1

SFR M et al. (2000) might not be exceptional if the SFR is near our 0.5 −1 0 estimate of 3.39 M yr . Wild et al. (2014) estimated 8– 10 9.5 9 8.5 8 7.5 7 −1 −1 20 M yr at 350 km s for the ionized outflow from NGC 0.5 inner-N hotspot 4676A. That the outflow in NGC 5394 is seen as absorption -1 0.4 far-N hotspot in the neutral line Na I, rather than in emission, might be yr SW disk 0.3 a feature of the later evolutionary stage, with the neutral Solar 0.2 outflow persisting longer, as suggested by the detection of

SFR M 0.1 neutral outflows in some dusty, red post-starburst galaxies 0 (e.g. Sato et al. 2009), and in recent gas-rich mergers (e.g. 10 9.5 9 8.5 8 7.5 7 Log Age yr NGC 34 in Schweizer & Seitzer 2007). Much higher spectral resolutions (R ≥ 5000) may be required to measure velocity Figure 29. Reconstructed SFR history, estimated from the Star- dispersions and disentangle the different inflow, outflow or burst model fits, of large (above) and and small (below) re- rotating components. gions of NGC 5394 as a function of look-back time, plotted in ∆(log[age]) = 0.3 intervals. Another feature of NGC 5394 is the high electron den- −3 sity of ne ∼ 750 cm in the central kpc, compared to the −3 central ne ∼ 200–300 cm more typical of star-forming at formation. As the results are rather noisy, mean SFRs are galaxies. This is not unique, e.g. Krabbe et al. (2014), in a estimated for wide ∆(log[age]) = 0.3 bins, by summing in Gemini Multi-Object Spectrograph study of 12 galaxies in each the (at-formation) masses of stars formed, and dividing 7 interacting pairs, found even lower [SII] 6717/6731 ratios −3 by the bin width in years. In this model the current stellar and higher ne (∼ 1000 cm ) in some examples, (such as AM 10 mass is only 2.44×10 M , reflecting the ‘lightweight’ IMF. 1054-325B and AN 1219-430), and concluded that there was The model also fits dust extinction to the stellar continuum, evidence interacting galaxies on average had higher ne but, Av = 1.06 mag for the nuclear region (notably less than the on the basis of moderate [OI]6300/Hα and [OIII]5007/Hβ emission-line AHα = 2.32 mag from the Balmer decrement), ratios, this was not generally caused by shock excitation. and Av = 0.60 and 0.31 for the SW and outer disk. Otherwise no cause was given except to attribute a central Fig 29 shows the resulting SFR histories. The nucleus increase in ne to gas inflow. The high central ne of NGC has a large old (∼ 10 Gyr) stellar population, plus resur- 5394 could be a consequence of sustained gas infall over gent star-formation (note increasing SFR) over the past 100 hundreds of Myr. Perhaps the ne enhancement is maximised Myr. The outer disk and tidal arms are dominated by in- for this particular interaction geometry (coplanar, prograde, termediate age (109–108 yr old) stars, the Hα-bright SW with larger companion) and stage. Large integral field sur- disk region has additional star-formation at age < 100 Myr, veys which can measure line ratios diagnostic of ne will be the star-formation in the far-N hotspot also goes back some able to quantify how it correlates with local and galaxy-scale 100 Myr, while the inner-N hotspot, with stronger Balmer properties. absorption, is mostly intermediate-age. The model fitted a large component of ∼ 0.9 Gyr age stars for the outer disk and tidal arms, suggesting the interaction with NGC 5395 8 CONCLUSIONS and enhanced star-formation have been going on for a long time, perhaps even with two previous close passages. (i) We examine CALIFA data for the spiral galaxy NGC The first perigalacticon could have triggered a pro- 5394, which is strongly interacting with the 4× more mas- longed influx of gas towards the nucleus, fuelling the sub- sive spiral NGC 5395. Most of the Hα emission from NGC sequent star-formation. If this gas was relatively low metal- 5394, and therefore the star-forming activity, is concentrated licity, the stellar metallicity would then decrease during the in the nucleus (75% in the central r ≤ 1 kpc) with addi- interaction, which (we argued in Section 5) could account for tional star-forming regions in the SW of the disk and in two the (model-fit) luminosity-weighted stellar metallicity being hotspots on the northern tidal arm. The galaxy is dusty, lower than the mass-weighted (Fig 20). Yet this influx did especially in the central r ≤ 1 kpc where the Hα/Hβ ratio not flatten the O3N2-metallicity gradient at r ≤ 10 arc- corresponds to a Hα extinction of 2.32 magnitudes. For the sec, perhaps because of competing re-enrichment by star- whole galaxy we estimate a dust-corrected, Hα based star- −1 formation (Torrey et al. 2012). Also O3N2 could be less sen- formation rate of 3.39 M yr (Chabrier IMF). The outer sitive to mixing of low-Z gas if prolonged star-formation and northern-arm hotspot has only 0.9% of the total SFR but reprocessing produced a high central N/O ratio (e.g. Belfiore the highest Hα equivalent width anywhere in the galaxy, et al. 2015). Tidally-induced gas infall may be be the cause over 100A.˚ Isolated, extranuclear concentrations of star- and the energy source of the relatively faint (a few percent formation, like this hotspot, are predicted to form in merger of the total line emission) but extensive (∼ 8 kpc diameter) simulations. outer ring of elevated N2, O3, [OI]/Hα and [SII]/Hα ratios, (ii) Much of the galaxy outside the inner disk has which are consistent with shocks (of velocity ∼ 250 km s−1). a post-starburst spectrum with strong Balmer absorption Extensive shock emission features similar to this this have (EWabs(Hδ) ≥ 6A).˚ The Hδ map and spectral model-fitting been found in other late-stage or strongly interacting, gas- show the galaxy underwent an extensive episode of rapid

c 2015 RAS, MNRAS 000, 1–?? 16 N.D. Roche, A. Humphrey, J.M. Gomes, P. Papaderos, P. Lagos, S.F. S´anchez star-formation 0.2–1.0 Gyr ago. Comparison with simula- ACKNOWLEDGMENTS tions and other mergers suggest this was triggered by the This study uses data provided by the Calar Alto Legacy In- earlier close passage of the two galaxies. tegral Field Area (CALIFA) survey (http://califa.caha.es/), (iii) NGC 5394 has a very high (for a purely star- funded by the Spanish Ministery of Science under grant forming galaxy) [NII]/Hα ratio of 0.63 at its centre. The ICTS-2009-10, and the Centro Astron´omico Hispano- [OIII]/Hβ, [OI]6300/Hα and [SII]/Hα ratios remain low Alem´an.Based on observations collected at the Centro As- here, with no central peak, and thus disfavour AGN and tron´omicoHispano Alem´an(CAHA) at Calar Alto, operated shocks as explanations. Instead the enhanced central N2 ra- jointly by the Max-Planck-Institut f¨urAstronomie and the tio is probably due to increased collisional excitation as- Instituto de Astrof´ısicade Andaluc´ıa(CSIC). NR, AH, JMG, PP and PL acknowledge Funda¸c˜ao sociated with the very high central electron density, ne ' 750 cm−3 on the basis of the [SII]6717/6731 ratio. para a Ciˆencia e a Tecnolog´ıa (FCT) support through UID/FIS/04434/2013, and through project FCOMP- (iv) There is an outer region, forming an irregular ring 01-0124-FEDER-029170 (Reference FCT PTDC/FIS- of pixels around the nucleus at 9 < r < 16 arcsec (2.25 < AST/3214/2012) funded by FCT-MEC (PIDDAC) and r < 4 kpc), with even higher [NII]/Hα ratios of 0.65–0.85, on FEDER (COMPETE), in addition to FP7 project average 0.74. Here the [OIII]/Hβ, [OI]6300/Hα and [SII]/Hα PIRSES-GA-2013-612701. NR acknowledges the support ratios are also high, suggesting a substantial (' 50% of Hα) of FCT postdoctoral grant SFRH/BI/52155/2013 followed contribution from shocks (we estimate, with v ∼ 250 kms−1 by CAUP2014-04UnI-BPD (University of Porto). AH and no pre-ionization). also acknowledges a Marie Curie Fellowship co-funded by the FP7 and the FCT (DFRH/WIIA/57/2011) (v) The O3N2 index of nebular metallicity with the and FP7 / FCT Complementary Support grant lower, Marino et al. (2013) calibration and the mass- SFRH/BI/52155/2013. PL is supported by a Postdoc- weighted model-fit stellar metallicity are similar at 12 + toral grant SFRH/BPD/72308/2010, funded by the FCT. log(O/H) ' 8.6 (0.76 solar). The metallicity from the JMG acknowledges support by the FCT through the O3N2 index shows a negative radial gradient of about -0.03 Fellowship SFRH/BPD/66958/2009 and POPH/FSE (EC) dex/kpc (typical of disk galaxies) at r < 10 arcsec (or one by FEDER funding through the program Programa Op- rhl), but this flattens at larger radii. Also the luminosity- eracional de Factores de Competitividade (COMPETE). weighted stellar metallicity is lower than the mass-weighted, PP is supported by FCT through the Investigador FCT especially in the outer disk, which might be the result of ear- Contract No. IF/01220/2013 and POPH/FSE (EC) by lier inflow of lower metallicity gas. FEDER funding through the program COMPETE.

(vi) The NaI(5890,5896) absorption in the spectrum of the nucleus has an equivalent width 6 to 7A,˚ at least 3 times stronger than expected from stellar absorption alone, and is REFERENCES blueshifted by 3A˚ over a > 1.5 kpc region. The residuals between the observed line and a stellar model indicate the Allen M.G., Groves B., Dopita M., Sutherland R., Kewley non-stellar absorption component is blueshifted by 223 km L.J., 2008, ApJS 178, 20. s−1. This is evidence for a neutral outflow of gas, at this Allende Prieto C., Lambert D.L., Asplund M., 2001, ApJ velocity and perpendicular to the disk plane, driven by the 556, L63. central starburst. Arp H., 1966, ApJS 14, 1.. (vii) Although the galaxy is viewed only ∼ 15◦ deg Arp H., 1969, A&A 3, 418. from face-on, a velocity gradient is visible in the Hα and Baldwin J.A., Phillips M.M., Terlevich R., 1981, PASP 93, [NII] emission lines, and in the stellar spectra (at least for 5. the V1200 data), consistent with a disk rotation velocity ∼ 156–180 km s−1 (as expected from the stellar mass). The Barnes J.E., 2004, MNRAS, 350, 798. Hα and [NII] velocity maps are very similar and show no sig- Barrera-Ballesteros J.K., et al., 2015, A&A 579, 45. nature of an ionized outflow (such as a blueshifted region) Belfiore F., et al. 2015, MNRAS 449, 867. either at the nucleus or the outer ring of high N2 ratios. This suggests that shocks in the outer high-excitation region Bordoloi R., et al., 2014, ApJ 794, 130. are not produced by the central outflow but more likely by Brinchmann J., Pettini M., Charlot, S., 2008, MNRAS 385, interaction-triggered gas inflow within the disk plane. 769. (viii) The strong neutral gas outflow, high electron den- Calzetti D., Armus L., Bohlin R.C., Kinney A.L., Koornneef −1 −2 sity and concentrated star formation ( > 1 M yr kpc ) at J., Storchi-Bergmann T., 2000, ApJ 533, 682. the nucleus, and the strong Balmer absorption and hotspots Cant´oJ., Elliott K.H., Meaburn J., Theokas A.C., 1980, elsewhere, are all a consequence of the prolonged interac- MNRAS 193, 911. tion with NGC 5395. Over a timescale as long as ∼ 1 Gyr and one or more perigalactica, this has reshaped the galaxy, Casoli F., et al., 1998, A&A, 331, 451. formed tidal arms, triggered starbursts and gas inflow, and Chen Yan-Mei, Tremonti C.A., Heckman T.M., Kauffmann now the SFR may be increasing again as the two galaxies G., Weiner B.J., Brinchmann J., Wang J., 2010, AJ 140, approach merger. 445.

c 2015 RAS, MNRAS 000, 1–?? CALIFA Spectroscopy: Interacting Galaxy NGC 5394 17

Chien Li-Hsin, Barnes J.E., Kewley L.J., Chambers K.C., J., Alonso-Herrero A., Garc´ıa-Mar´ın M., 2010, A&A 517, 2007, ApJ 660, 105. 28. Cid Fernandes R., et al. 2013, A&A 557, 86. Papaderos P., et al., 2013 A&A 555, 1. Cortese L., et al., 2014, ApJ 795, 37. Pettini M., Pagel B.E.J. 2004, MNRAS, 348, L59. Farage C.L., McGregor P.J., Dopita M.A., Bicknell G.V., Puerari I., Valdez-Guti´errezM., Hern´andez-L´opez I., 2005, 2010, ApJ 724, 267. AJ 130, 1524. Garc´ıa-BenitoR., et al., 2015, A&A, 576, 135. Renaud F., Bournaud F., Duc P-A., 2015, MNRAS, 466, 2038. Garc´ıa-LorenzoB., et al. 2015, A&A, 573, 59. Rich J.A., Kewley L.J., Dopita M.A., 2011, ApJ, 734, 87. Gomes J.M., et al., 2015, A&A, submitted. Rich J.A., Kewley L.J., Dopita M.A., 2014, ApJ 781, L12. Goto T., 2008, MNRAS 381, 187. Rich, J.A., Torrey P., Kewley L.J., Dopita M.A., Rupke Heckman T.M., Armus L., Weaver K.A., Wang J., 1999, ApJ D.S.N., 2012, ApJ, 753, 5. 517, 130. Rodr´ıguez-Baras M., Rosales-Ortega F.F., D´ıaz A.I., Heckman T.M., Lehnert M.D., Strickland D.K., Armus L., S´anchez S.F., Pasquali A., 2014, MNRAS 442, 495. 2000, ApJS, 129, 493. Rupke D.S.N., Kewley L.J., Barnes J.E., 2010, ApJ 710, 156. Higdon S.J.U, Higdon J. L.,Smith B. J.,Hancock M., 2014, ApJ 787, 103. S´anchez S.F., et al. 2012, A&A 538, 8. Ho I-Ting, Kudritzki R-P, Kewley L.J., Jabran Zahid H., S´anchez S.F., et al. 2014, A&A 563, 49. Dopita M.A., Bresolin F., Rupke. D.S.N., 2015, MNRAS S´anchez S.F., et al. 2015, A&A 574, 47. 448, 2030. S´anchez-Blazquez P., et al. 2014, A&A 570, 6. Howell J.H., et al., 2010, ApJ 715, 572. Sato T., Martin C.L., Noeske K.G., Koo D.C., Lotz J.M., Husemann B., et al., 2013, A&A 549, 87 2009, ApJ 696, 214. Kaneko H., Kuno N., Iono D., Tamura Y., Tosaki T.; Nakan- Schweizer F., Seitzer P., 2007, AJ 133, 2132. ishi K., Sawada T., 2013, PASJ 65, 20. Sobral D., Best P.N., Smail I., Mobasher B., Stott J., Nisbet Kauffmann, G. et al. 2003, MNRAS, 346, 1055 D., 2014, MNRAS 437, 3516. Kaufman M., Brinks E., Elmegreen B.G., Elmegreen D.M.,; Struck C., Kaufman M., Brinks E., Thomasson M., Klari´cM., Struck C., Thomasson M., Vogel S., 1999, AJ 118, Elmegreen B.G., Elmegreen D.M., 2005, MNRAS 364, 69. 1577. Struck C., Smith, B.J., 2012, MNRAS, 422, 2444. Kaufman M., Sheth K., Struck C.; Elmegreen B.G., Smith, B.J., Soria R., Struck C., Giroux M.L., Swartz D. A.; Thomasson M., Elmegreen D. M., Brinks E., 2002, AJ 123, Yukita M., 2014, AJ 147, 60. 702. Smith B.J., Struck C., Hancock M., Appleton P.N., Char- Kaufman M., Grupe D., Elmegreen B.G. Elmegreen D.M., mandaris V., Reach W.T., 2007, AJ 133, 791. Struck C., Brinks E., 2012, AJ 144, 156. Torrey P., Cox T.J., Kewley L., Hernquist L., 2012, ApJ Keel W.C., Kennicutt Jr. R.C., Hummel E., van der Hulst 746, 108. J.M., 1985, AJ 90, 108. Walcher C J., et al. 2014, A&A 569, 1. Kennicutt Jr. R. C., 1998, ARA&A 36, 189. Wild V.; Groves B., Heckman T., Sonnentrucker P., Armus Kewley L.J., Dopita M.A., Leitherer C., Dav´eR., Yuan T., L., Schiminovich D., Johnson B., Martins L., Lamassa S., Allen M., Groves B., Sutherland R., 2013, ApJ, 774, 100. 2011, MNRAS 410, 1593. Krabbe A.C., Rosa D.A., Dors O.L., Pastoriza M.G., Winge Wild V., Rosales-Ortega F., Falcon-Barroso J, et al., 2014, C., H¨ageleG.F., Cardaci M.V., Rodrigues I., 2014, MNRAS A&A 567, 143. 437, 1155. Yuan T.-T., Kewley L.J., Swinbank A.M., Richard J., 2012, Lanz L., et al., 2013, ApJ 768, 90. ApJ 759, 66. Le Tiran L., Lehnert M.D., van Driel W., Nesvadba N.P.H., Di Matteo P., 2011, A&A 534, 4. Lehmer B.D., Alexander D.M., Bauer F.E., Brandt W.N., Goulding A.D., Jenkins L.P., Ptak A., Roberts T.P., 2010, ApJ, 724, 559. L´opez-Hern´andez J., Terlevich E., Terlevich R., Rosa- Gonz´alezD., Daz .; Garc´ıa-BenitoR.,;V´ılchez J., H¨ageleG., 2013, MNRAS 430, 472. Marino R.A., et al., 2013, A&A 559, 114. Monreal-Ibero A., Arribas S., Colina L., Rodrguez-Zaur´ın

c 2015 RAS, MNRAS 000, 1–??